The task is to show that $\lambda*|\mu|=\epsilon$, can you please look at my proof and show, why the Dirichlet convolution of $2$ multiplicative functions is multiplicative ?

Let $n=p_1^{a_1}\cdot p_2^{a_2}\dots\cdot p_r^{a_r}$, and $\lambda(n)=(-1)^{a_1+a_2+\dots a_r}$ and

$\mu(n)=\begin{cases}1&\text{if}\ n=1\\(-1)^{t}&\text{if}n=p_1\cdot p_2\dots p_t, \ \text{where $p_i$ are distinct primes}\\0&\text{if $\exists p$ such that $p^2\mid n$}\end{cases}$

I have to show that the Dirichlet convolution $\lambda*|\mu|=\epsilon$ where $\epsilon$ is the Dirichlet convolution identity, i.e. $\epsilon(1)=\begin{cases}1&\text{if $n=1$}\\0&\text{else}\end{cases}$

Now if I define $F(n):=\left(\lambda*|\mu|\right)(n)$, and assume that $F$ is multiplicative then I think I can prove it;

If $\underline{n=1}$:

$\displaystyle F(1)=\left(\lambda*|\mu|\right)(1)=\sum\limits_{d\mid 1\atop n\in\mathbb N}\lambda(1)|\mu(1)|=(-1)^0|1|=1$

If $\underline{n=p^k}$:

$\displaystyle F(p^k)=\left(\lambda*|\mu|\right)(p^k)=\sum\limits_{d\mid p^k\atop n\in\mathbb N}\lambda(d)|\mu(\frac{p^k}{d})|$

$=\lambda(1)\underbrace{|\mu(p^k)|}_{=0}+\lambda(p)\underbrace{|\mu(p^{k-1})|}_{=0}+\dots+\underbrace{\lambda(p^{k-1})}_{=(-1)^{k-1}} \underbrace{|\mu(p)|}_{|-1|=1}+\underbrace{\lambda(p^k)}_{(-1)^k}\underbrace{|\mu(1)|}_{=1}$


So If $F$ is multiplicative then;

$F(n)=F(p_1^{a_1})F(p_2^{a_2})\dots F(p_r^{a_r})=0\cdot0\dots\cdot0=0$

do you agree ?


F is multiplicative, as the Dirichlet convolution of two multiplicative arithmetic functions, take $\mu$ and $\nu$ , multiplicative arithmetic functions, then their convolution is the function defined by $\sum_{d|mn} \mu(d) \nu(mn/d)$, but taking m and n for which $(m,n) = 1$, we could devide the set of all deviders of $mn$ two into two sets, having just intersection 1, the set of dividors of m and the set of all dividors of n, so the sum becomes $\sum_{d|m}\mu(d)\nu(m/d)\sum_{d_1|n}\mu(d_1)\nu(n/d_1)$, as everything is finite we have that we can use termwise multiplication by no restrictions. Proof is straightforward.

  • $\begingroup$ other proof : multiplicative Dirichlet series have an Euler product, and the product of two Euler products is still and Euler product. CQFD $\endgroup$ – reuns Jan 9 '16 at 17:30
  • $\begingroup$ "We could divide the set of all dividers of $mn$ into two sets, having just intersection $1$." Is that true? Isn't $(3, 4) = 1$ where $6 | 12$ but $6$ is neither a divisor of $3$ nor $4$. $\endgroup$ – gowrath Feb 9 '17 at 4:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.